On Thu, 18 Sep 2025 09:00:28 +0200 Paul Gevers <[email protected]> wrote: > Hi, > > On 9/17/25 13:42, Jeremy Bícha wrote: > >> Did normaliz break the ABI? Or why does pynormaliz's tests fail when > >> compiled > >> against old normaliz but run against the newer version? > > > It seems that without an answer to the above the following: > > > A maintainer upload > > would also get the current situation unstuck. > > > would be equally papering over the real problem as an binNMU would. We > prefer the real problem being fixed. E.g. there might be third party > software out there where a SONAME bump in src:normaliz makes it clear a > rebuild is needed, while papering over doesn't.
I believe the underlying problem is a bug when using pynormaliz with normaliz 3.10.5 -- see [1]. It's fixed when running pynormaliz against a build of the current master branch of normaliz, so hopefully things should be back to normal once normaliz 3.11.0 is released and packaged in a few weeks. [1] https://github.com/Normaliz/PyNormaliz/issues/122
